1 /* stolen_chunk_of_toke.c - from perl 5.8.8 toke.c
3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
4 * 2000, 2001, 2002, 2003, 2004, 2005, 2006, by Larry Wall and others
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
12 * "It all comes from here, the stench and the peril." --Frodo
16 * this is all blatantly stolen. I sincerely hopes it doesn't fuck anything
17 * up but if it does blame me (Matt S Trout), not the poor original authors
20 /* the following #defines are stolen from assorted headers, not toke.c (mst) */
22 #define skipspace(a) S_skipspace(aTHX_ a, 0)
23 #define peekspace(a) S_skipspace(aTHX_ a, 1)
24 #define skipspace_force(a) S_skipspace(aTHX_ a, 2)
25 #define incline(a) S_incline(aTHX_ a)
26 #define filter_gets(a,b,c) S_filter_gets(aTHX_ a,b,c)
27 #define scan_str(a,b,c) S_scan_str(aTHX_ a,b,c)
28 #define scan_word(a,b,c,d,e) S_scan_word(aTHX_ a,b,c,d,e)
29 #define scan_ident(a,b,c,d,e) S_scan_ident(aTHX_ a,b,c,d,e)
31 STATIC void S_incline(pTHX_ char *s);
32 STATIC char* S_skipspace(pTHX_ char *s, int incline);
33 STATIC char * S_filter_gets(pTHX_ SV *sv, PerlIO *fp, STRLEN append);
34 STATIC char* S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims);
35 STATIC char* S_scan_word(pTHX_ char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp);
37 #define DPTR2FPTR(t,p) ((t)PTR2nat(p)) /* data pointer to function pointer */
38 #define FPTR2DPTR(t,p) ((t)PTR2nat(p)) /* function pointer to data pointer */
39 #define PTR2nat(p) (PTRV)(p) /* pointer to integer of PTRSIZE */
41 /* conditionalise these two because as of 5.9.5 we already get them from
44 #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)safemalloc((MEM_SIZE)((n)*sizeof(t)))))
47 #define SvPVX_const(sv) ((const char*) (0 + SvPVX(sv)))
49 #ifndef MEM_WRAP_CHECK_
50 #define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t),
53 #define SvPV_renew(sv,n) \
54 STMT_START { SvLEN_set(sv, n); \
55 SvPV_set((sv), (MEM_WRAP_CHECK_(n,char) \
56 (char*)saferealloc((Malloc_t)SvPVX(sv), \
60 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
62 /* On MacOS, respect nonbreaking spaces */
63 #ifdef MACOS_TRADITIONAL
64 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
66 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
70 * Normally, during compile time, PL_curcop == &PL_compiling is true. However,
71 * Devel::Declare makes the interpreter call back to perl during compile time,
72 * which temporarily enters runtime. Then perl space calls various functions
73 * from this file, which are designed to work during compile time. They all
74 * happen to operate on PL_curcop, not PL_compiling. That doesn't make a
75 * difference in the core, but it does for Devel::Declare, which operates at
76 * runtime, but still wants to mangle the things that are about to be compiled.
77 * That's why we define our own PL_curcop and make it point to PL_compiling
81 #define PL_curcop (&PL_compiling)
83 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
85 #define LEX_NORMAL 10 /* normal code (ie not within "...") */
86 #define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
87 #define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
88 #define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
89 #define LEX_INTERPSTART 6 /* expecting the start of a $var */
91 /* at end of code, eg "$x" followed by: */
92 #define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
93 #define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
95 #define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
96 string or after \E, $foo, etc */
97 #define LEX_INTERPCONST 2 /* NOT USED */
98 #define LEX_FORMLINE 1 /* expecting a format line */
99 #define LEX_KNOWNEXT 0 /* next token known; just return it */
101 /* and these two are my own madness (mst) */
103 #if PERL_REVISION == 5 && PERL_VERSION == 8 && PERL_SUBVERSION >= 8
104 #define PERL_5_8_8_PLUS
107 #if PERL_REVISION == 5 && PERL_VERSION > 8
108 #define PERL_5_9_PLUS
111 #if !defined(PERL_5_9_PLUS) && defined(PERL_IMPLICIT_CONTEXT)
112 /* These two are not exported from the core on Windows. With 5.9+
113 it's not an issue, because they're part of the PL_parser structure,
114 which is exported. On multiplicity/thread builds we can work
115 around the lack of export by this formulation, where we provide
116 a substitute implementation of the unexported accessor functions.
117 On single-interpreter builds we can't, because access is directly
118 via symbols that are not exported. */
119 # define Perl_Ilinestart_ptr my_Ilinestart_ptr
120 char **my_Ilinestart_ptr(pTHX) { return &(aTHX->Ilinestart); }
121 # define Perl_Isublex_info_ptr my_Isublex_info_ptr
122 static SUBLEXINFO *my_Isublex_info_ptr(pTHX) { return &(aTHX->Isublex_info); }
126 /* 5.9+ moves a bunch of things to a PL_parser struct so we need to
127 declare the backcompat macros for things to still work (mst) */
129 /* XXX temporary backwards compatibility */
130 #define PL_lex_brackets (PL_parser->lex_brackets)
131 #define PL_lex_brackstack (PL_parser->lex_brackstack)
132 #define PL_lex_casemods (PL_parser->lex_casemods)
133 #define PL_lex_casestack (PL_parser->lex_casestack)
134 #define PL_lex_defer (PL_parser->lex_defer)
135 #define PL_lex_dojoin (PL_parser->lex_dojoin)
136 #define PL_lex_expect (PL_parser->lex_expect)
137 #define PL_lex_formbrack (PL_parser->lex_formbrack)
138 #define PL_lex_inpat (PL_parser->lex_inpat)
139 #define PL_lex_inwhat (PL_parser->lex_inwhat)
140 #define PL_lex_op (PL_parser->lex_op)
141 #define PL_lex_repl (PL_parser->lex_repl)
142 #define PL_lex_starts (PL_parser->lex_starts)
143 #define PL_lex_stuff (PL_parser->lex_stuff)
144 #define PL_multi_start (PL_parser->multi_start)
145 #define PL_multi_open (PL_parser->multi_open)
146 #define PL_multi_close (PL_parser->multi_close)
147 #define PL_pending_ident (PL_parser->pending_ident)
148 #define PL_preambled (PL_parser->preambled)
149 #define PL_sublex_info (PL_parser->sublex_info)
150 #define PL_linestr (PL_parser->linestr)
151 #define PL_sublex_info (PL_parser->sublex_info)
152 #define PL_linestr (PL_parser->linestr)
153 #define PL_expect (PL_parser->expect)
154 #define PL_copline (PL_parser->copline)
155 #define PL_bufptr (PL_parser->bufptr)
156 #define PL_oldbufptr (PL_parser->oldbufptr)
157 #define PL_oldoldbufptr (PL_parser->oldoldbufptr)
158 #define PL_linestart (PL_parser->linestart)
159 #define PL_bufend (PL_parser->bufend)
160 #define PL_last_uni (PL_parser->last_uni)
161 #define PL_last_lop (PL_parser->last_lop)
162 #define PL_last_lop_op (PL_parser->last_lop_op)
163 #define PL_lex_state (PL_parser->lex_state)
164 #define PL_rsfp (PL_parser->rsfp)
165 #define PL_rsfp_filters (PL_parser->rsfp_filters)
166 #define PL_in_my (PL_parser->in_my)
167 #define PL_in_my_stash (PL_parser->in_my_stash)
168 #define PL_tokenbuf (PL_parser->tokenbuf)
169 #define PL_multi_end (PL_parser->multi_end)
170 #define PL_error_count (PL_parser->error_count)
171 #define PL_nexttoke (PL_parser->nexttoke)
172 /* these are from the non-PERL_MAD path but I don't -think- I need
173 the PERL_MAD stuff since my code isn't really populating things (mst) */
175 # define PL_curforce (PL_parser->curforce)
176 # define PL_lasttoke (PL_parser->lasttoke)
178 # define PL_nexttype (PL_parser->nexttype)
179 # define PL_nextval (PL_parser->nextval)
181 /* end of backcompat macros from 5.9 toke.c (mst) */
184 /* when ccflags include -DDEBUGGING we need this for earlier 5.8 perls */
185 #ifndef SvPV_nolen_const
186 #define SvPV_nolen_const SvPV_nolen
189 /* utf8_to_uvchr_buf() not defined in earlier perls, but less-capable
190 * substitute is available */
192 #ifndef utf8_to_uvchr_buf
193 #define utf8_to_uvchr_buf(s, e, lp) ((e), utf8_to_uvchr(s, lp))
196 #ifndef isIDFIRST_lazy_if_safe
197 # define isIDFIRST_lazy_if_safe(p,e,UTF) \
198 ((! UTF || p > e) ? isIDFIRST_lazy_if(p,UTF) : 0)
200 #ifndef isALNUM_lazy_if_safe
201 # define isALNUM_lazy_if_safe(p,e,UTF) \
202 ((! UTF || p > e) ? isALNUM_lazy_if(p,UTF) : 0)
204 #ifndef isALNUM_utf8_safe
205 # define isALNUM_utf8_safe(p,e) ((p > e) ? isALNUM_utf8(p) : 0)
208 /* and now we're back to the toke.c stuff again (mst) */
210 static const char ident_too_long[] =
211 "Identifier too long";
212 static const char c_without_g[] =
213 "Use of /c modifier is meaningless without /g";
214 static const char c_in_subst[] =
215 "Use of /c modifier is meaningless in s///";
217 #ifdef USE_UTF8_SCRIPTS
218 # define UTF (!IN_BYTES)
220 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
223 /* Invoke the idxth filter function for the current rsfp. */
224 /* maxlen 0 = read one text line */
226 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
231 if (!PL_rsfp_filters)
233 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
234 /* Provide a default input filter to make life easy. */
235 /* Note that we append to the line. This is handy. */
236 DEBUG_P(PerlIO_printf(Perl_debug_log,
237 "filter_read %d: from rsfp\n", idx));
241 const int old_len = SvCUR(buf_sv);
243 /* ensure buf_sv is large enough */
244 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
245 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
246 if (PerlIO_error(PL_rsfp))
247 return -1; /* error */
249 return 0 ; /* end of file */
251 SvCUR_set(buf_sv, old_len + len) ;
254 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
255 if (PerlIO_error(PL_rsfp))
256 return -1; /* error */
258 return 0 ; /* end of file */
261 return SvCUR(buf_sv);
263 /* Skip this filter slot if filter has been deleted */
264 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
265 DEBUG_P(PerlIO_printf(Perl_debug_log,
266 "filter_read %d: skipped (filter deleted)\n",
268 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
270 /* Get function pointer hidden within datasv */
271 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
272 DEBUG_P(PerlIO_printf(Perl_debug_log,
273 "filter_read %d: via function %p (%s)\n",
274 idx, datasv, SvPV_nolen_const(datasv)));
275 /* Call function. The function is expected to */
276 /* call "FILTER_READ(idx+1, buf_sv)" first. */
277 /* Return: <0:error, =0:eof, >0:not eof */
278 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
282 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
284 #ifdef PERL_CR_FILTER
285 if (!PL_rsfp_filters) {
286 filter_add(S_cr_textfilter,NULL);
289 if (PL_rsfp_filters) {
291 SvCUR_set(sv, 0); /* start with empty line */
292 if (FILTER_READ(0, sv, 0) > 0)
293 return ( SvPVX(sv) ) ;
298 return (sv_gets(sv, fp, append));
303 * Called to gobble the appropriate amount and type of whitespace.
304 * Skips comments as well.
308 S_skipspace(pTHX_ register char *s, int incline)
310 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
311 while (s < PL_bufend && SPACE_OR_TAB(*s))
317 SSize_t oldprevlen, oldoldprevlen;
318 SSize_t oldloplen = 0, oldunilen = 0;
319 while (s < PL_bufend && isSPACE(*s)) {
320 if (*s++ == '\n' && ((incline == 2) || (PL_in_eval && !PL_rsfp && !incline)))
325 if (s < PL_bufend && *s == '#') {
326 while (s < PL_bufend && *s != '\n')
330 if (PL_in_eval && !PL_rsfp && !incline) {
337 /* also skip leading whitespace on the beginning of a line before deciding
338 * whether or not to recharge the linestr. --rafl
340 while (s < PL_bufend && isSPACE(*s)) {
341 if (*s++ == '\n' && PL_in_eval && !PL_rsfp && !incline)
345 /* only continue to recharge the buffer if we're at the end
346 * of the buffer, we're not reading from a source filter, and
347 * we're in normal lexing mode
349 if (s < PL_bufend || !PL_rsfp || PL_lex_inwhat ||
350 PL_lex_state == LEX_FORMLINE)
353 /* try to recharge the buffer */
354 if ((s = filter_gets(PL_linestr, PL_rsfp,
355 (prevlen = SvCUR(PL_linestr)))) == Nullch)
357 /* end of file. Add on the -p or -n magic */
360 ";}continue{print or die qq(-p destination: $!\\n);}");
361 PL_minus_n = PL_minus_p = 0;
363 else if (PL_minus_n) {
364 sv_setpvn(PL_linestr, ";}", 2);
368 sv_setpvn(PL_linestr,";", 1);
370 /* reset variables for next time we lex */
371 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
373 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
374 PL_last_lop = PL_last_uni = Nullch;
376 /* In perl versions previous to p4-rawid: //depot/perl@32954 -P
377 * preprocessors were supported here. We don't support -P at all, even
378 * on perls that support it, and use the following chunk from blead
382 /* Close the filehandle. Could be from
383 * STDIN, or a regular file. If we were reading code from
384 * STDIN (because the commandline held no -e or filename)
385 * then we don't close it, we reset it so the code can
386 * read from STDIN too.
389 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
390 PerlIO_clearerr(PL_rsfp);
392 (void)PerlIO_close(PL_rsfp);
397 /* not at end of file, so we only read another line */
398 /* make corresponding updates to old pointers, for yyerror() */
399 oldprevlen = PL_oldbufptr - PL_bufend;
400 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
402 oldunilen = PL_last_uni - PL_bufend;
404 oldloplen = PL_last_lop - PL_bufend;
405 PL_linestart = PL_bufptr = s + prevlen;
406 PL_bufend = s + SvCUR(PL_linestr);
408 PL_oldbufptr = s + oldprevlen;
409 PL_oldoldbufptr = s + oldoldprevlen;
411 PL_last_uni = s + oldunilen;
413 PL_last_lop = s + oldloplen;
417 /* debugger active and we're not compiling the debugger code,
418 * so store the line into the debugger's array of lines
420 if (PERLDB_LINE && PL_curstash != PL_debstash) {
421 AV *fileav = CopFILEAV(PL_curcop);
423 SV * const sv = NEWSV(85,0);
424 sv_upgrade(sv, SVt_PVMG);
425 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
428 av_store(fileav,(I32)CopLINE(PL_curcop),sv);
435 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
437 register char *d = dest;
438 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
441 Perl_croak(aTHX_ ident_too_long);
442 if (UTF && isIDFIRST_utf8_safe((const U8*) s, (const U8*) PL_bufend)) {
443 /* The UTF-8 case must come first, otherwise things
444 * like c\N{COMBINING TILDE} would start failing, as the
445 * isALNUM case below would gobble the 'c' up.
448 char *t = s + UTF8SKIP(s);
449 while (isIDCONT_utf8_safe((const U8*) t, (const U8*) PL_bufend)) {
453 Perl_croak(aTHX_ "%s", ident_too_long);
454 Copy(s, d, t - s, char);
458 else if (isALNUM(*s))
461 } while (isWORDCHAR_A(*s) && d < e);
464 && isIDFIRST_lazy_if_safe(s+1, PL_bufend, UTF))
470 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
484 * This subroutine has nothing to do with tilting, whether at windmills
485 * or pinball tables. Its name is short for "increment line". It
486 * increments the current line number in CopLINE(PL_curcop) and checks
487 * to see whether the line starts with a comment of the form
488 * # line 500 "foo.pm"
489 * If so, it sets the current line number and file to the values in the comment.
493 S_incline(pTHX_ char *s)
500 CopLINE_inc(PL_curcop);
503 while (SPACE_OR_TAB(*s)) s++;
504 if (strnEQ(s, "line", 4))
508 if (SPACE_OR_TAB(*s))
512 while (SPACE_OR_TAB(*s)) s++;
518 while (SPACE_OR_TAB(*s))
520 if (*s == '"' && (t = strchr(s+1, '"'))) {
525 for (t = s; !isSPACE(*t); t++) ;
528 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
530 if (*e != '\n' && *e != '\0')
531 return; /* false alarm */
536 /* this chunk was added to S_incline during 5.8.8. I don't know why but I don't
537 honestly care since I probably want to be bug-compatible anyway (mst) */
539 /* ... my kingdom for a perl parser in perl ... (mst) */
541 #ifdef PERL_5_8_8_PLUS
543 const char *cf = CopFILE(PL_curcop);
544 if (cf && strlen(cf) > 7 && strnEQ(cf, "(eval ", 6)) {
545 /* must copy *{"::_<(eval N)[oldfilename:L]"}
546 * to *{"::_<newfilename"} */
547 char smallbuf[256], smallbuf2[256];
548 char *tmpbuf, *tmpbuf2;
550 STRLEN tmplen = strlen(cf);
551 STRLEN tmplen2 = strlen(s);
552 if (tmplen + 3 < sizeof smallbuf)
555 Newx(tmpbuf, tmplen + 3, char);
556 if (tmplen2 + 3 < sizeof smallbuf2)
559 Newx(tmpbuf2, tmplen2 + 3, char);
560 tmpbuf[0] = tmpbuf2[0] = '_';
561 tmpbuf[1] = tmpbuf2[1] = '<';
562 memcpy(tmpbuf + 2, cf, ++tmplen);
563 memcpy(tmpbuf2 + 2, s, ++tmplen2);
565 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
567 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
569 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
570 /* adjust ${"::_<newfilename"} to store the new file name */
571 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
572 GvHV(gv2) = (HV*)SvREFCNT_inc(GvHV(*gvp));
573 GvAV(gv2) = (AV*)SvREFCNT_inc(GvAV(*gvp));
575 if (tmpbuf != smallbuf) Safefree(tmpbuf);
576 if (tmpbuf2 != smallbuf2) Safefree(tmpbuf2);
580 /* second endif closes out the "are we 5.8.(8+)" conditional */
581 CopFILE_free(PL_curcop);
582 CopFILE_set(PL_curcop, s);
585 CopLINE_set(PL_curcop, atoi(n)-1);
589 takes: start position in buffer
590 keep_quoted preserve \ on the embedded delimiter(s)
591 keep_delims preserve the delimiters around the string
592 returns: position to continue reading from buffer
593 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
594 updates the read buffer.
596 This subroutine pulls a string out of the input. It is called for:
597 q single quotes q(literal text)
598 ' single quotes 'literal text'
599 qq double quotes qq(interpolate $here please)
600 " double quotes "interpolate $here please"
601 qx backticks qx(/bin/ls -l)
602 ` backticks `/bin/ls -l`
603 qw quote words @EXPORT_OK = qw( func() $spam )
604 m// regexp match m/this/
605 s/// regexp substitute s/this/that/
606 tr/// string transliterate tr/this/that/
607 y/// string transliterate y/this/that/
608 ($*@) sub prototypes sub foo ($)
609 (stuff) sub attr parameters sub foo : attr(stuff)
610 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
612 In most of these cases (all but <>, patterns and transliterate)
613 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
614 calls scan_str(). s/// makes yylex() call scan_subst() which calls
615 scan_str(). tr/// and y/// make yylex() call scan_trans() which
618 It skips whitespace before the string starts, and treats the first
619 character as the delimiter. If the delimiter is one of ([{< then
620 the corresponding "close" character )]}> is used as the closing
621 delimiter. It allows quoting of delimiters, and if the string has
622 balanced delimiters ([{<>}]) it allows nesting.
624 On success, the SV with the resulting string is put into lex_stuff or,
625 if that is already non-NULL, into lex_repl. The second case occurs only
626 when parsing the RHS of the special constructs s/// and tr/// (y///).
627 For convenience, the terminating delimiter character is stuffed into
632 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
634 SV *sv; /* scalar value: string */
635 char *tmps; /* temp string, used for delimiter matching */
636 register char *s = start; /* current position in the buffer */
637 register char term; /* terminating character */
638 register char *to; /* current position in the sv's data */
639 I32 brackets = 1; /* bracket nesting level */
640 bool has_utf8 = FALSE; /* is there any utf8 content? */
641 I32 termcode; /* terminating char. code */
642 /* 5.8.7+ uses UTF8_MAXBYTES but also its utf8.h defs _MAXLEN to it so
643 I'm reasonably hopeful this won't destroy anything (mst) */
644 U8 termstr[UTF8_MAXLEN]; /* terminating string */
645 STRLEN termlen; /* length of terminating string */
646 char *last = NULL; /* last position for nesting bracket */
648 /* skip space before the delimiter */
652 /* mark where we are, in case we need to report errors */
655 /* after skipping whitespace, the next character is the terminator */
658 termcode = termstr[0] = term;
662 termcode = utf8_to_uvchr_buf((U8*)s, PL_bufend, &termlen);
663 Copy(s, termstr, termlen, U8);
664 if (!UTF8_IS_INVARIANT(term))
668 /* mark where we are */
669 PL_multi_start = CopLINE(PL_curcop);
670 PL_multi_open = term;
672 /* find corresponding closing delimiter */
673 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
674 termcode = termstr[0] = term = tmps[5];
676 PL_multi_close = term;
678 /* create a new SV to hold the contents. 87 is leak category, I'm
679 assuming. 79 is the SV's initial length. What a random number. */
681 sv_upgrade(sv, SVt_PVIV);
682 SvIV_set(sv, termcode);
683 (void)SvPOK_only(sv); /* validate pointer */
685 /* move past delimiter and try to read a complete string */
687 sv_catpvn(sv, s, termlen);
690 if (PL_encoding && !UTF) {
694 int offset = s - SvPVX_const(PL_linestr);
695 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
696 &offset, (char*)termstr, termlen);
697 const char *ns = SvPVX_const(PL_linestr) + offset;
698 char *svlast = SvEND(sv) - 1;
700 for (; s < ns; s++) {
701 if (*s == '\n' && !PL_rsfp)
702 CopLINE_inc(PL_curcop);
707 /* handle quoted delimiters */
708 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
710 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
712 if ((svlast-1 - t) % 2) {
716 SvCUR_set(sv, SvCUR(sv) - 1);
721 if (PL_multi_open == PL_multi_close) {
729 for (t = w = last; t < svlast; w++, t++) {
730 /* At here, all closes are "was quoted" one,
731 so we don't check PL_multi_close. */
733 if (!keep_quoted && *(t+1) == PL_multi_open)
738 else if (*t == PL_multi_open)
746 SvCUR_set(sv, w - SvPVX_const(sv));
755 SvCUR_set(sv, SvCUR(sv) - 1);
761 /* extend sv if need be */
762 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
763 /* set 'to' to the next character in the sv's string */
764 to = SvPVX(sv)+SvCUR(sv);
766 /* if open delimiter is the close delimiter read unbridle */
767 if (PL_multi_open == PL_multi_close) {
768 for (; s < PL_bufend; s++,to++) {
769 /* embedded newlines increment the current line number */
770 if (*s == '\n' && !PL_rsfp)
771 CopLINE_inc(PL_curcop);
772 /* handle quoted delimiters */
773 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
774 if (!keep_quoted && s[1] == term)
776 /* any other quotes are simply copied straight through */
780 /* terminate when run out of buffer (the for() condition), or
781 have found the terminator */
782 else if (*s == term) {
785 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
788 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
794 /* if the terminator isn't the same as the start character (e.g.,
795 matched brackets), we have to allow more in the quoting, and
796 be prepared for nested brackets.
799 /* read until we run out of string, or we find the terminator */
800 for (; s < PL_bufend; s++,to++) {
801 /* embedded newlines increment the line count */
802 if (*s == '\n' && !PL_rsfp)
803 CopLINE_inc(PL_curcop);
804 /* backslashes can escape the open or closing characters */
805 if (*s == '\\' && s+1 < PL_bufend) {
807 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
812 /* allow nested opens and closes */
813 else if (*s == PL_multi_close && --brackets <= 0)
815 else if (*s == PL_multi_open)
817 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
822 /* terminate the copied string and update the sv's end-of-string */
824 SvCUR_set(sv, to - SvPVX_const(sv));
827 * this next chunk reads more into the buffer if we're not done yet
831 break; /* handle case where we are done yet :-) */
833 #ifndef PERL_STRICT_CR
834 if (to - SvPVX_const(sv) >= 2) {
835 if ((to[-2] == '\r' && to[-1] == '\n') ||
836 (to[-2] == '\n' && to[-1] == '\r'))
840 SvCUR_set(sv, to - SvPVX_const(sv));
842 else if (to[-1] == '\r')
845 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
850 /* if we're out of file, or a read fails, bail and reset the current
851 line marker so we can report where the unterminated string began
854 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
856 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
859 /* we read a line, so increment our line counter */
860 CopLINE_inc(PL_curcop);
862 /* update debugger info */
863 if (PERLDB_LINE && PL_curstash != PL_debstash) {
864 AV *fileav = CopFILEAV(PL_curcop);
866 SV *sv = NEWSV(88,0);
867 sv_upgrade(sv, SVt_PVMG);
868 sv_setsv(sv,PL_linestr);
871 av_store(fileav, (I32)CopLINE(PL_curcop), sv);
875 /* having changed the buffer, we must update PL_bufend */
876 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
877 PL_last_lop = PL_last_uni = Nullch;
880 /* at this point, we have successfully read the delimited string */
882 if (!PL_encoding || UTF) {
884 sv_catpvn(sv, s, termlen);
887 if (has_utf8 || PL_encoding)
890 PL_multi_end = CopLINE(PL_curcop);
892 /* if we allocated too much space, give some back */
893 if (SvCUR(sv) + 5 < SvLEN(sv)) {
894 SvLEN_set(sv, SvCUR(sv) + 1);
895 /* 5.8.8 uses SvPV_renew, no prior version actually has the damn thing (mst) */
896 #ifdef PERL_5_8_8_PLUS
897 SvPV_renew(sv, SvLEN(sv));
899 Renew(SvPVX(sv), SvLEN(sv), char);
903 /* decide whether this is the first or second quoted string we've read
914 #define XFAKEBRACK 128
917 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
921 char *bracket = Nullch;
927 e = d + destlen - 3; /* two-character token, ending NUL */
929 while (isDIGIT(*s)) {
931 Perl_croak(aTHX_ ident_too_long);
938 Perl_croak(aTHX_ ident_too_long);
939 if (UTF && isIDFIRST_utf8_safe((const U8*) s, (const U8*) PL_bufend)) {
940 /* The UTF-8 case must come first, otherwise things
941 * like c\N{COMBINING TILDE} would start failing, as the
942 * isALNUM case below would gobble the 'c' up.
945 char *t = s + UTF8SKIP(s);
946 while (isIDCONT_utf8_safe((const U8*) t, (const U8*) PL_bufend)) {
950 Perl_croak(aTHX_ "%s", ident_too_long);
951 Copy(s, d, t - s, char);
955 else if (isALNUM(*s))
958 } while (isWORDCHAR_A(*s) && d < e);
959 else if (*s == '\'' && isIDFIRST_lazy_if_safe(s+1,send,UTF)) {
964 else if (*s == ':' && s[1] == ':') {
975 if (PL_lex_state != LEX_NORMAL)
976 PL_lex_state = LEX_INTERPENDMAYBE;
979 if (*s == '$' && s[1] &&
980 ( isALNUM_lazy_if_safe(s+1,send,UTF)
981 || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
989 /* we always call this with ck_uni == 0, so no need for check_uni() */
995 if (*d == '^' && *s && isCONTROLVAR(*s)) {
1000 if (isSPACE(s[-1])) {
1002 const char ch = *s++;
1003 if (!SPACE_OR_TAB(ch)) {
1009 if (isIDFIRST_lazy_if_safe(d,d+destlen,UTF)) {
1014 && isIDFIRST_utf8_safe(e, send))
1018 while (e < send && isIDFIRST_utf8_safe(e, send))
1021 Copy(s, d, e - s, char);
1026 while ((isALNUM(*s) || *s == ':') && d < e)
1029 Perl_croak(aTHX_ ident_too_long);
1032 while (s < send && SPACE_OR_TAB(*s)) s++;
1033 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
1034 /* we don't want perl to guess what is meant. the keyword
1035 * parser decides that later. (rafl)
1038 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
1039 const char *brack = *s == '[' ? "[...]" : "{...}";
1040 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
1041 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
1042 funny, dest, brack, funny, dest, brack);
1046 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
1050 /* Handle extended ${^Foo} variables
1051 * 1999-02-27 mjd-perl-patch@plover.com */
1052 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
1056 while (isALNUM(*s) && d < e) {
1060 Perl_croak(aTHX_ ident_too_long);
1065 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
1066 PL_lex_state = LEX_INTERPEND;
1071 /* we don't want perl to guess what is meant. the keyword
1072 * parser decides that later. (rafl)
1075 if (PL_lex_state == LEX_NORMAL) {
1076 if (ckWARN(WARN_AMBIGUOUS) &&
1077 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
1079 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
1080 "Ambiguous use of %c{%s} resolved to %c%s",
1081 funny, dest, funny, dest);
1087 s = bracket; /* let the parser handle it */
1091 /* don't intuit. we really just want the string. (rafl) */
1093 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
1094 PL_lex_state = LEX_INTERPEND;